We explored whether the proteomic analysis of exhaled breath condensate (EBC) may provide biomarkers for non-invasive screening for the early detection of lung cancer (LC). EBC was collected from 192 individuals (49 control (C), 49 risk factor-smoking (S), 46 chronic obstructive pulmonary disease (COPD) and 48 LC (LC)). Using LC-MS/MS, 348 different proteins with a different pattern among the four groups were identified in EBC samples. Significantly more proteins were identified in the EBC from LC compared to other groups (C:12.4±1.3; S:15.3±1; COPD:14±1.6; LC:24.2±3.6; p=0.0001). Furthermore, the average number of proteins identified per sample was significantly higher in LC patients and ROC curve analysis showed an area under the curve of 0.8, indicating diagnostic value. Proteins frequently detected in EBC, such as dermcidin and hornerin, along with others much less frequently detected, such as hemoglobin and histones, were identified. Cytokeratins (KRTs) were the most abundant proteins in EBC samples and levels of KRT6A, KRT6B and KRT6C isoforms were significantly higher in samples from LC patients (p=0.0031, p=0.0011 and p=0.0009, respectively). Moreover, the amount of most KRTs in EBC samples from LC patients showed a significant positive correlation with tumor size. Finally, we used a random forest algorithm to generate a robust model using EBC protein data for the diagnosis of patients with LC where the area under the ROC curve obtained indicated a good classification (82%). Thus, this study demonstrates that the proteomic analysis of EBC samples is an appropriated approach to develop biomarkers for the diagnosis of lung cancer.